Little Lesson 1: Playing for the Endgame

Grandmaster Siegbert Tarrasch (1862-1934) famously said, "Before the endgame, the gods have placed the middlegame." But that doesn't mean that one shouldn't concern himself with the coming endgame while the middlegame is still going on. While it is true that one is often concerned with slash-and-burn tactics in the middlegame, so that considerations of endgame pawn structure or of endgame piece advantages take a back seat to king-hunting, there are still many times when one can plan ahead for the endgame while still playing the middlegame. The following game, from the third round of the West Chester Chess Club's 2007 championship tournament, provides an example.

White: Hans Hermans 1 e4 2 d4 3 Nc3 4 Nf3

Black: Keith Brian Johnson d6 Nf6 g6

Not as aggressive as the Austrian Attack (4 f4), but hard for Black to get play against.

5 Be2 6 O-O 7 Bf4 8 Bg3 9 fxg3

Bg7 c6 O-O Nh5 Nxg3

At the non-master level, it might not be clear why this is wrong; the game is young, and it might seem that having the half-open f-file on which to attack is no small matter. The pawn structure on move nine might not resemble the endgame pawn structure very closely, so the crippling of White's pawns might not seem very important--in many a swashbuckling game, after all, pawn structure is less immediately important than the king-hunt. According to the Club's resident master, though, it is clear that this is a mistake: In such positions the opening of the f-file almost never sufficiently compensates for the crippling of White's pawns. As will be seen, taking with the f-pawn will create a crippled three-on-two pawn majority for White, meaning that White will not be able to force the creation of a passed pawn and that Black will be able to stop White's three pawns with his own two. The main point I want to make now is this: If you're playing the Black side in such a position, you should notice the crippled pawn island and, even at this early juncture, you should think about how to turn it into an endgame win.

9 10 Bc4 11 dxe5 12 Nxe5 13 Qf3

Nd7 e5 Nxe5 Bxe5 Be6

The Club master points out that Black had 13 ... Qb6+ 14 Kh1 Qxb2 15 Ne2 Be6 (15 Bxf7+ Kg7 will result in White's bishop's being pinned against his queen, and Black will win the bishop with 16 Ne2 Be6).

14 Bxe6

fxe6

Now look at the pawns. Black has a healthy five-on-four pawn majority on the queenside and in the center, while White's three-on-two pawn majority on the kingside is crippled. Without help from any piece, Black can force a passed pawn on the queenside or in the center, while White cannot force a passed pawn on the kingside without piece help. (On the other hand, if White's kingside pawns were on the g- and f-files, as they would be had White recaptured with his h-pawn, White could make a passed pawn, if he got his forward g-pawn to g4 before Black put his h-pawn on h5; despite doubling the g-pawn, taking with the h-pawn would not have resulted in the same crippling of the kingside pawn island that taking with the f-pawn resulted in. So, on move nine, White shouldn't think, "Well, I'll double my g-pawn either way, and if I take with the f-pawn I'll at least get a half-open f-file for my rooks." Taking with the f-pawn is definitely worse for the endgame than is taking with the h-pawn--it creates a permanent, structural weakness that can be exploited in the endgame, in a way that taking with the h-pawn does not.) If the game reaches a king-and-pawn ending with this pawn structure, Black will win. So, if Black has the opportunity, he should steer the game toward a king-and-pawn ending. He might not have the chance; there's still plenty of opportunity for middlegame play, and White may be expected not to cooperate with Black's plans. But Black has an advantage White doesn't have: He can feel free to trade pieces if he wants to. White has to try to keep pieces on the board, unless he can nullify Black's potential endgame advantage.

15 Qg4

Rf6

Again, the Club master points out 15 ... Qb6+, with 16 Kh1 17 Qxb2 Qxe6+ 18 Kh8 Ne2 19 Qxc2 to follow.

16 Rxf6

Qxf6

Here there is a small trap: If White moves his rook to f1, attacking Black's queen, then ... Bd4+ will force the White king to the corner, after which ... Qxf1 will be checkmate. White avoids the trap.

17 Qf3

This is smart! It gives Black the chance to fix White's pawn structure for him by exchanging queens on f3. Black doesn't oblige. But, at least, White has given his opponent the chance to go wrong. A good practical rule of thumb is to give your opponent choices, and to give him the opportunity to choose badly. He might not oblige. But he might. On the other hand, it's not a good idea to go out of your way to give your opponent such choices--if he chooses well, you shouldn't have hurt your game by giving him the chance to go wrong in the first place, unless you're in such desperate straits that you have no other practical choice.

17 18 Qxf6

Rf8 Rxf6

Again inviting Rf1, with the subsequent Bd4+ and mate.

19 Ne2

d5

After the game, Hans pointed out the possibility of 19 ... Bxb2, winning a pawn, since 20 Rb1 Ba3 21 Rxb7? Bc5+ mates by 22 Kh1 Rf1+ 23 Ng1 Rxg1#. I simply missed it, continuing with my plan of exploiting my healthy pawn majority by playing 19 ... d5.

20 exd5

exd5

Even with pieces still on the board, Black is winning; both the pawn structure and the bishop-vs.-knight advantage favor him. (Bishops, because of their greater mobility, are generally better than knights in endgames when there are pawns on both sides of the board. With pawns on only one side of the board, knights are sometimes better, since they can control squares of either color but bishops cannot.) But with pieces on the board, it's always possible for White to defend actively and to give Black maximal difficulty; since the king-and-pawn ending is won and is simple even for a non-master, Black hopes to trade pieces. Conversely, White, in order to mount the most active defense possible, should try to keep the rooks on the board.

21 c3 22 Rf1

Bc7

This simplifies Black's life. Now, should he keep the bishop and knight on the board, since the bishop is better than the knight? Well, he could; but why do that, when the king-and-pawn ending is won? Make life easy on yourself.

22 23 Nd4 24 cxd4 25 Kxf1

Bb6+ Bxd4+ Rxf1+

After this, it's a sure win for Black. Black might not have perfect endgame technique, and he might misplay it into a theoretical draw at some point; but Black should win this without undue difficulty. The whole game has been about simplifying into a won king-and-pawn ending. Both kings will move toward the center; Black's two kingside pawns will stop White's crippled threesome; Black will force a passed pawn on the queenside; White's king will be forced to give up ground and to move backward and let the passed pawn advance; and then Black will be able to walk away and pick off other White pawns while White's king is busy taking the passed pawn, enabling Black to win. The first passed pawn created, the one that will result from the four-on-three queenside majority, will end up being a decoy!

25 26 Ke2 27 Ke3 28 b4

Kf7 Ke6 b6 h6

I'm told this was unnecessary. Its idea was to make it impossible for the White king to walk in and attack Black's kingside pawns via g4. Even if it was unnecessary, the idea of forming a wall of squares that the enemy king can't cross is sometimes important in endgames and is one to remember.

29 g4 30 Kd3 31 a3 32 dxc5+

Kd6 a5 c5 bxc5

After 32 ... bxc5

After 37 Kb4 (analysis)

You should convince yourself that this is OK for Black. Neither 33 b5 a4 (preventing 34 a4 and controlling the square b3), when White's king can't attack either the Black a-pawn or the Black central pawns and Black can take the time to pick off the White b-pawn, nor 33 bxa5 Kc6, when White's king again can't attack the Black central pawns and Black can take the time to pick off the White a-pawns, works for White. Even 33 b5 c4+ 34 Kd4 remains a win for Black, provided that he plays 34 ... a4, preventing White's own a4. If 35 g3 Kc7 36 Kc3 Kb6 37 Kb4 (analysis diagram), Black plays 37 ... c3, which is followed by 38 Kxc3 Kxb5 39 Kd4 Kc6, winning much as in the game. Only if Black allows White to play both b5 and a4 does the win evaporate. But 33 b5 a4 seems simpler than 33 b5 c4+, because the White king remains walled off from any useful activity by the a-, c-, and d-pawns, and the Black king may eat White's b-pawn at its leisure.

33 bxc5+ 34 Kc3 35 Kd3

Kxc5 d4+ a4

Black does have to be careful not to fall into something like the following (see Tempo Moves): 36 Kd2 Kc4 37 Kc2 d3+ 38 Kd2 Kb3 39 Kxd3 Kxa3 40 Kc3, when White's king pins Black's king on the rook file--if 40 ... Ka2, then 41 Kc2. If White had no kingside pawn moves--if the pawns were already locked, and if the g2-pawn had already advanced as far as possible--then ... a3 would give Black an extra tempo, forcing White's king to c1; Black's king could then escape via b3 and walk to the kingside and munch pawns. (Without the pawns, it would still be a draw, even if White's king escaped to b3, since king-and-rook-pawn vs. king is drawn when the defending king can reach the relevant corner.) But if White still had a kingside pawn move available, then ... a3 wouldn't help release Black's king from its prison on the a-file; White would simply make a pawn move, and then all that would be left would be White's king's shuffling back and forth on a2 and a1 while White's king did the same on c2 and c1. It would be a draw! (It's important that White could make the last pawn move on the kingside, forcing Black to shuffle his king back and forth. You can assure yourself that he could.) Black must not fall into such a variation. (And White must save a pawn move in case Black does fall into such a variation.)

36 g3 37 h3 38 h4

Kd5 Kc5 Kd5

When White runs out of pawn moves, his king will have to retreat.

39 h5

g5

Black's two pawns stop White's three pawns: White's crippled pawn majority comes to nothing. This is why, earlier in the game, Black could look forward to a won ending. (39 g5 hxg5 would also have left White's pawns stopped.) Now White has no pawn moves, so Black would win even in such a variation as given above, with ... a3 providing the winning tempo.

40 Kd2

Kc4

Perhaps 40 ... Ke4, headed toward the kingside pawns, would be a little more obviously winning.

41 Kc2 42 Kd2 43 Kd1

d3+ Kd4 Kc3

Again, perhaps 43 ... Ke3 would be a little more obviously winning. Black simply had the win with the a-pawn calculated.

44 Kc1 45 Kd2 46 Kxd3

Kb3 Kxa3 Kb2

And now the a-pawn will queen; Black won in thirteen more routine moves. Please notice that without the a-pawns, the game would still have been won for Black, as he would have walked his king toward the kingside pawns upon abandoning his d-pawn, as with the perhaps-more-obvious 43 ... Ke3.

As can be seen, there are times when one may steer a game toward a favorable endgame, and there are times, when one side or the other has a long-term, structural weakness (rather than a short-term, dynamic weakness), when middlegame moves are made with an eye toward reaching a favorable endgame--or, reciprocally, with an eye toward not reaching an unfavorable one. The gods may have placed the middlegame before the endgame, but they have not also forced us to make the transition from one to the other blindly.